Artmarket.com plans to roll out Artprice Visual Search, an AI-powered image search tool designed to help collectors and art professionals identify and explore artworks through visual similarities and historical market data. The company says a global beta will open to subscribers in early December 2026, with commercial access scheduled for January 2027 through higher-priced Premium and Professional subscriptions. The technology combines computer vision with Artprice’s art-market database, drawing on decades of cataloguing and documentation. Its launch highlights a specialized AI use case: connecting visual recognition with provenance, artist information and auction records rather than treating an artwork as an image alone.
Artprice Visual Search Uses AI to Connect Artwork Images With Market History
Finding an artwork through a photograph is becoming a more sophisticated problem than matching colors, shapes or visual patterns. Collectors and art professionals may also need to understand who created a work, its provenance, artistic context, materials, auction history and relationship to other pieces. Artmarket.com believes the combination of visual recognition and structured art-market records can make that process more useful.
The company has announced Artprice Visual Search, a proprietary AI-powered search system scheduled to enter global beta testing in early December 2026. A commercial release is planned for January 2027, when the feature will become part of Artprice’s higher-priced Premium and Professional subscriptions. The December beta is intended to give existing subscribers an opportunity to test the technology before the paid rollout.
The announcement marks an expansion of Artprice’s existing data and analytics business into image-based discovery. Rather than relying only on text queries such as artist names, movements or auction categories, the system is designed to let users search through an artwork’s visual characteristics and retrieve related works from Artprice’s database.
Combining Computer Vision With Art-Historical Metadata
Artmarket.com says the system was trained on 23 million annotated artwork reproductions, with each image associated with as many as 180 structured documentary fields. These include provenance, artistic school, technique, materials, dimensions, signatures, iconography, catalogue raisonné references and auction records.
The company says its wider image collection contains nearly 145 million reproductions, which are being processed for the platform. It also reports that its Artprice databases cover more than 930,000 artists and include auction traceability dating back to 1700.
The distinction between the training dataset and the broader image collection is important. The 23 million annotated reproductions form the documented core used to train the system, while the nearly 145 million images represent the larger archive being processed. Artmarket.com has not independently published a technical evaluation showing how accuracy varies across artists, periods, image quality or works outside its documented collection.
According to the company, Artprice Visual Search combines computer vision with a retrieval-augmented generation (RAG) architecture that cross-references an image’s visual characteristics against historical and documentary records. Its intended output is a list of 49 related works, each assigned a relevance percentage. Matches can reflect the same artist, subject matter, composition, formal structure, color relationships or artistic technique.
That approach targets a limitation of image-only recognition: visually similar works are not necessarily connected by authorship, provenance or historical context. Adding structured records could make search results more informative, particularly when users are researching works that share an aesthetic vocabulary but differ substantially in market value or attribution.
However, Artmarket.com’s claims that the system delivers exceptionally high identification accuracy and reduces visual uncertainty to a statistically marginal level have not been accompanied in the announcement by independent benchmark results, error rates or comparative testing. Those measurements will matter to professional users, especially when search results inform valuation or acquisition decisions.
A Subscription Strategy Tied to AI Features
The rollout also has a commercial objective. Artmarket.com plans to make the beta available to its subscriber base during December, then reserve the commercial feature for Premium and Professional customers from January 2027. The company says the higher-end subscriptions will carry significantly higher prices.
This is a direct attempt to turn proprietary AI capabilities into higher average revenue per user (ARPU). The company expects the beta period to encourage existing customers to upgrade, although it has not disclosed the planned subscription prices, conversion targets or expected revenue contribution from Visual Search.
The strategy reflects a wider software-market pattern: specialized AI features are increasingly being packaged into premium tiers rather than offered as standalone tools. For Artprice, the commercial case rests on whether visual discovery offers enough additional value to collectors, dealers, researchers and auction-market professionals to justify a more expensive subscription.
A Digital Art Market With Room for Better Discovery
The addressable market is substantial, but its economics are not straightforward. The Art Basel and UBS Global Art Market Report 2026, authored by Arts Economics, estimates that global art sales increased 4% to $59.6 billion in 2025, with approximately 41.5 million transactions. The recovery followed two years of declining market values, and performance remained uneven across regions and price segments.
In that environment, better discovery tools could help users navigate a large and fragmented supply of artworks. Collectors may use visual search to explore unfamiliar artists or identify works that fit a particular aesthetic direction. Dealers and researchers could use it to investigate comparable works, while auction professionals may find image-based discovery useful alongside conventional catalogue and artist searches.
Artmarket.com also positions Visual Search as part of a broader AI portfolio that includes Intuitive Artmarket and Blind Spot, its existing analytical and predictive systems. The company’s longer-term ambition is to use its historical records and image archive to identify relationships and emerging artistic patterns that may be difficult to detect through conventional search.
The technology’s success will ultimately depend on more than the size of its image collection. Independent accuracy testing, transparent relevance scoring, image-rights practices, provenance quality and the ability to distinguish visual resemblance from verified attribution will be central to professional adoption.
Artprice Visual Search is a notable move toward domain-specific AI for the art market. Its December beta and January subscription launch will provide the first opportunity for customers to assess whether the combination of computer vision and decades of art documentation can deliver a meaningful improvement over conventional art-market search.
Market Landscape
Artprice Visual Search enters the intersection of computer vision, vertical AI, visual discovery and art-market intelligence.
The opportunity is shaped by several market developments:
- Visual-first discovery: Collectors increasingly need ways to explore works through composition, color, subject, medium and style rather than relying exclusively on artist names.
- Domain-specific AI: Specialized systems can combine image recognition with curated industry data, potentially delivering more context than general-purpose image search.
- Digitization of art-market records: Historical auction records, provenance information, catalogue references and image archives create opportunities for data-driven research.
- Subscription monetization: Art-market platforms can package specialized search and analytical capabilities into higher-value professional services.
- Trust and attribution: Visual similarity is not equivalent to verified authorship, provenance or authenticity. Professional applications require clear distinctions between algorithmic matches and documented facts.
The Art Basel and UBS Global Art Market Report 2026 estimates that global art-market sales reached $59.6 billion in 2025, up 4% year over year, with approximately 41.5 million transactions.
This creates a substantial potential audience for digital discovery and research tools, although the market’s recovery remains uneven. Artprice’s potential differentiation is its combination of visual search with a large historical database, rather than image recognition alone.
Top Insights
- Artprice Visual Search combines computer vision with historical art-market metadata, targeting image-based discovery beyond conventional artist-name and keyword searches.
- Artmarket.com says the system was trained on 23 million annotated artwork reproductions, with 180 documentary fields associated with each reproduction.
- The company is processing a broader archive of nearly 145 million artwork images, although this is distinct from the reported training dataset.
- Search results are designed to return 49 visually or historically related works, each with a relevance percentage.
- The January 2027 commercial rollout will place Visual Search in higher-priced Premium and Professional subscriptions, supporting Artmarket.com’s ARPU expansion strategy.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
